How the management of sepsis has improved and will improve further

Introduction

The management of sepsis in hospitals is significantly better today than it was 30 years ago. However, sepsis-associated mortality rates still remain unacceptably high, and we believe that we must work together to embrace new strategies in order to improve patient outcomes still further. The recent improvement in outcomes has been characterised by the successive introduction of multiple interventions and therapies, and is an ongoing process. We believe that the current wave of clinical trial data relating to a number of new interventions should be viewed in the context of this trend towards ever-improving management of the condition. In this paper, we will draw comparisons with the management of acute myocardial infarction (AMI), which illustrates a similar pattern of incremental improvements in therapy.

AMI is rightly considered to be one of the scourges of the late 20^th and early 21^st centuries, and is currently the leading cause of death worldwide (1,2). Due to high mortality rates and substantial healthcare costs, AMI has been the focus of global media and public attention, and extensive research efforts and considerable resources have been directed towards its prevention. This has resulted in steady improvements in the management of AMI and a subsequent reduction in AMI-related mortality. There are lessons to learn from AMI for those of us engaged in the fight against sepsis.

Both sepsis and AMI are acute conditions that kill rapidly, and both require a combination of therapeutic approaches, since no single therapy is available that is capable of targeting the multiple pathophysiological components of either condition. Furthermore, achieving significant mortality reductions in sepsis patient populations today seems as fraught with difficulties as similar reductions in AMI-associated mortality seemed 40 years ago.

Sepsis-associated mortality is high

Sepsis is an aggressive and multifactorial disease state that has been ranked as the tenth-leading cause of death in the US (3). Data from the 1995 US Census (4) showed that the condition accounted for approximately 236,000 deaths annually. Moreover, the actual number of deaths associated with the condition may be even higher than current estimates suggest. Patients usually die of sepsis during the course of an underlying disease, and deaths are often attributed to these conditions rather than to sepsis (5-7).

The unacceptably high death toll from sepsis may be attributed to its increasing incidence together with the high, and increasing, risk of death and severe organ dysfunction associated with this condition. Placebo-controlled trials, which provide the best benchmarks for assessing changes in mortality as new interventions are developed, suggest an overall mortality rate associated with sepsis of 28–50% at 28 days (8-12). Such benchmarks are important when one considers that mortality rates of up to 80–90% have been observed in some institutions(5, 6, 8, 13-15). This persistent, high mortality rate ranks sepsis alongside some of the biggest acute killers in hospital, namely AMI (16-20), stroke (21), and ruptured abdominal aortic aneurysm (22, 23) (see Table 1).

Table 1. Mortality rates due to sepsis, AMI, stroke and ruptured abdominal aortic aneurysm from clinical trial data^{8–12,16–23}

* 28-day mortality.
† 30-day in-hospital mortality rates indicated by literature review of clinical trial data.
‡ 14-day mortality amongst patients receiving standard care treatment in the International Stroke Trial.
§ Operative mortality rates.

In addition to the high rate of mortality associated with sepsis, the emergence of a number of key factors mean that this condition poses an ever more serious threat to ever more people. In most developed countries, an aging population is being sustained by increasingly sophisticated medical technologies — and sepsis-induced mortality has been shown to increase with increasing age; there are more immunocompromised patients; there is increasing bacterial resistance to antibiotics; and there is the increasing prevalence of acquired immune deficiency syndrome (AIDS) (4-6, 14, 24-28).

Against this backdrop, it is a tribute to the successful development and implementation of treatment strategies that the overall mortality rate in sepsis has, to some degree, been reduced. A recent meta-analysis revealed only a modest decrease in septic shock-induced mortality over the last 30 years, despite the successful implementation of therapeutic strategies; the authors suggested that this might be explained by a worsening risk profile amongst affected patients (25). Other data indicate that sepsis-induced mortality rates have remained the same or increased (6, 15). Data also predict that the overall incidence of sepsis is increasing by approximately 1.5% per year (4). It thus appears that, without ongoing improvements in the way sepsis is managed, death rates from sepsis will escalate and the success of other life-saving therapies, such as chemotherapy and transplantation surgery, will continue to be undermined.

Incremental mortality reductions: the AMI example

In the 1960s, in-hospital AMI-associated mortality rates averaged approximately 25–30% (17). Whilst death rates remain consistently high (~50%) in those patients who do not receive timely medical care, an overview of the literature shows that current 30-day in-hospital mortality rates range from 2.7% to 9.6% (16-20).

This reduction in mortality was achieved by the development and implementation of successive life-saving interventions that have included the use of aspirin, fibrinolytics, and angiotensin-converting enzyme (ACE)-inhibitors, to name but a few (see Table 2). The benefits of each were demonstrated in clinical trials yielding important, but modest, mortality reductions. In the GUSTO (Global Use of Strategies to Open Occluded Coronary Arteries) I trial, for example, the benefit of accelerated tissue plasminogen activator (tPA) over streptokinase amounted to a significant, but less than 1%, absolute reduction in mortality (29). Over the course of many years, however, the additive effect of many such incremental advances has been immense.

Developments in the management of sepsis should be regarded in the same way as those in AMI, and never more so than in the current era when, for the first time, studies have reported distinct mortality reductions with new interventions. 

Implementation

Of course, the development of new and effective therapeutic strategies is worthless unless they are implemented. To this end, the increasing application of evidence-based medicine is proving invaluable, and again the example of AMI is instructive.

Some of the largest and most sophisticated clinical trials in modern medicine have been conducted for the management of AMI. Trials like GUSTO, ISIS (International Study of Infarct Survival), TIMI (Thrombolysis In Myocardial Infarction), and most recently HOPE (Heart Outcomes Prevention Evaluation) are recognised by physicians around the globe. Such trials have driven an evidence-based approach in the management of AMI with the introduction of guidelines for all aspects of AMI — from the early recognition and initiation of treatment in the ambulance and emergency room, to reperfusion strategies and secondary prevention with the use of aspirin, statins and ACE-inhibitors. Indeed, recent data indicate that the adoption of evidence-based approaches has considerably increased the use of pharmacologic therapies in hospital, thus representing a significant contribution to declining mortality rates post-AMI (30).

To aid this progression, committees representing experts in the field and the key societies (including the American College of Cardiology [ACC], the American Heart Association [AHA], and the European Society of Cardiology [ESC]) were formed to agree on achievable goals and standard procedures for patient management.

Table 2. Current strategies and therapies for the management of AMI patients in hospital^{17, 31–33}

The actual implementation of proven AMI treatment strategies was, however, quite slow, especially at first. In the sepsis arena, more rapid uptake of new therapies must be aimed for, and in this respect the AIDS example is encouraging. The high public and political profile of AIDS in the 1980s and early 1990s enabled the provision of significant resource funds and the rapid evolution of therapies for this disease. The causative agent of AIDS was identified in 1984 and by the mid 1990s highly effective combination antiretroviral treatment strategies were in place, enabling significant improvements in disease management to be made (34-39).

Thus, by raising public and political awareness of sepsis and by adopting evidence-based approaches, we hope to accelerate the implementation of new therapies for its management, and achieve significant, incremental reductions in mortality.

Achieving mortality reductions in sepsis

We are reaching a critical point in the management of sepsis patients. Important studies have been completed in recent years that have identified successful evidence-based therapeutic and disease management strategies for critically ill patients, including those with sepsis.

This research has expanded our understanding of the biochemistry of sepsis, improved definitions of sepsis, and enabled more rapid identification of sepsis patients and more successful treatment of underlying infections (40). Most importantly, for the first time, therapies have been developed that have shown consistent, positive effects on mortality.

Currently, available strategies for the management of sepsis patients include: timely patient identification and diagnosis; rapid identification of causative organisms; appropriate, timely antimicrobial therapy; improved ventilatory techniques (low-pressure ventilation); appropriate (goal-directed) haemodynamic support; targeted pharmacological therapies (drotrecogin alfa [activated]), immunological therapy, and glycaemic control (intensive insulin therapy); appropriate nutrition; effective supportive therapies (prophylaxis against stress ulcers, administration of anticoagulants, and dialysis); and patient management by highly qualified clinicians and nursing staff (6, 10, 26, 28, 40-48).

These strategies have helped to reduce the incidence of infections, support failing organs and prevent complications (40, 43). They demonstrate that reductions in mortality are achievable, and that an incremental, combination therapy-based approach is the key to diminishing sepsis-associated mortality.

The following sections summarise the most recent developments in sepsis patient management strategies. Whilst we cannot address the problem of sicker patient populations, through the continued refinement and widespread adoption of such strategies, and the introduction of novel therapies, mortality rates may decline still further, in line with those achieved in AMI patients.

General strategies

Haemodynamic support

Haemodynamic therapy, in the form of rapid fluid resuscitation, should be commenced to reverse hypotension, hypovolemia and organ dysfunction, and to restore effective tissue perfusion and cellular metabolism in patients with sepsis and severe sepsis. If fluid resuscitation fails to restore adequate arterial pressure and organ perfusion, vasopressors, and potentially inotropes, should be administered (26, 51). Care must be taken when commencing patients on vasopressor or inotropic therapy, since studies have indicated that their administration causes detrimental effects, including: impaired splanchnic blood flow and oxygenation; reduced pH; and negative effects on growth and thyroid hormones (49, 52, 53).

Antibacterial therapy

Timely administration of antibacterial therapy is crucial for the management of sepsis patients 6, 45. Approximately 10% of all sepsis patients do not receive prompt antibiotic therapy and face a mortality rate 10–15% higher than those receiving timely therapies (28).

Since the common approach of initiating broad-spectrum antibiotic coverage promotes the development of resistant bacterial strains and reduces the number of available antimicrobial agents, it is suggested that the following strategies should be considered to maximize antibacterial efficacy: risk stratification of patients; restrictions on certain antibiotic drug classes; scheduled changes in antibiotics; combination antibiotic therapies; antibiotic rotation; and infection control practices (28, 40, 45).

Evaluation of the adequacy of source control in the critically ill patient can be difficult. As with other modes, of anti-infective therapy, effective source control measures are expected to result in clinical improvement (62).

Low-pressure ventilation

An estimated 25–42% of sepsis patients develop acute respiratory distress syndrome (ARDS), which requires mechanical ventilation (MV) (26). Despite developments over the last decade, the mortality rate associated with ARDS may exceed 40% (49).  Older approaches to MV employed high tidal volumes (10–15 ml/kg ideal body weight [IBW]), which have been associated with stretch-induced lung injury and mortality rates of 14.3–48% (50). The use of low tidal volumes (6 ml/kg/IBW) has been shown to reduce the risk of mortality in ARDS patients by 22% and increase the number of ventilator-free days in the first 28 days (12 vs. 10; p=0.007) compared to conventional treatment (42).

Furthermore, whilst the benefits of low-volume ventilation are now widely accepted, many patients are still ventilated at high tidal volumes, resulting in unnecessary injury and mortality. It is hoped that the wider dissemination of study data and the development of standardised ventilatory strategies will reduce the rates of mortality associated with its use in sepsis patients.

Glycaemic control

Hyperglycaemia associated with insulin resistance is common in critically ill patients and may lead to severe infections, polyneuropathy, multiple organ failure and death (46). In a recent study of surgical ICU patients, intensive insulin therapy reduced in-hospital mortality by 34%, and yielded reductions in the following conditions: incidence of acute renal failure (41%); episodes of septicaemia (46%); incidence of bloodstream infections (46%); abnormal levels of inflammatory markers; prolonged use of antibiotics; median number of red cell transfusions required (50%); risk of polyneuropathy (44%); length of stay in the ICU (16.8% of patients in the intensive treatment group stayed for over 5 days vs. 26.3% of control patients); and requirements for prolonged mechanical ventilation (46). In a comparable Belgian study, intensive insulin therapy reduced morbidity and mortality in critically ill surgical patients by 40% (59). Finally, data indicate that the maintenance of normoglycaemia with insulin improves outcomes in critically ill patients and may reduce treatment costs (46, 60).

Dialysis

The incidence of renal failure requiring dialysis is low in sepsis patients (<5%), however, the risk of mortality in sepsis patients with acute renal failure (ARF) exceeds 50% (26). Renal replacement therapies in critically ill patients are primarily limited to intermittent renal replacement therapy (IRRT) and continuous renal replacement therapy (CRRT). Considerable debate surrounds the management of ARF patients and whether the more costly CRRT should be adopted. A recent meta-analysis suggested that CRRT was associated with a significant mortality reduction (p<0.01) when adjustments were made for study quality and severity of illness (48).

Sepsis-specific strategies

Early goal-directed therapy

Rivers et al. (2001) (10) recently showed that the early provision of therapy to maintain haemodynamic goals significantly reduced in-hospital mortality in severe sepsis and septic shock patients compared to standard haemodynamic therapy (30.5% vs. 46.5%; p=0.009). Significant reductions in 28-day (33.3%vs. 49.2%; p=0.01) and 60-day (44.3% vs. 56.9%; p=0.03) mortality rates were achieved and hospital length of stay was significantly reduced (18.4±15.0 vs. 14.6±14.5 days; p=0.04).

Recombinant Human Activated Protein C (rhAPC)

rhAPC represents the first therapy against sepsis to show efficacy in a Phase III clinical trial (8). This treatment exerts its affects by multiple modes of action: inhibition of coagulation; reduced inflammation; and promotion of fibrinolysis (28, 54, 55).

In the large international Phase III PROWESS (recombinant human Activated Protein C Worldwide Evaluation in Severe Sepsis) trial (n=1690), administration of rhAPC to patients with severe sepsis resulted in 6% absolute reductions in 28-day all-cause mortality compared with control patients (24.7% vs. 30.8%; p=0.005) (8). Furthermore, a subgroup analysis of efficacy data stratified by APACHE (Acute Physiology and Chronic Health Evaluation) II score showed a 13% absolute mortality reduction (31% vs. 44% with placebo) in patients with a high risk of death (APACHE II score >=25) (55) and questioned efficacy in the lowest APACHE II quartile.

In addition to the significant mortality reductions observed, treatment with rhAPC resulted in improvements in cardiovascular and respiratory organ dysfunction in the first seven days of treatment (p<0.05) (56). Mortality reductions were achieved at the expense of a modest 1.4% increase in the risk of serious bleeding (8).

Immunological therapy - low-dose corticosteroids

Despite doubts cast on the efficacy of corticosteroid therapy by trials in the 1980s, studies over the last decade have indicated that, in low doses, steroids reduce morbidity and mortality in septic shock patients(11, 12, 44, 47, 57, 58). These studies support the role of stress dose steroids in reducing mortality in septic shock patients(57, 58).  A French multicentre trial (n=300) demonstrated that survival of patients with septic shock is significantly improved when low-doses of hydrocortisone and fludrocortisone are applied(57). This 28-day survival benefit was seen only in patients with relative adrenal insufficiency (non-responders to the corticotrophin test). In non-responders, there were 73 deaths (63%) in the placebo group and 60 deaths (53%) in the corticosteroid group (p=0.02). Vasopressor therapy was withdrawn within 28 days in 46 patients (40%) in the placebo group and in 65 patients (57%) in the corticosteroid group (p=0.001). No increase of adverse events in the treatment group was found (57). 

In conclusion, administration of stress dose steroids may play a significant role in the management of select patients with septic shocK (47, 57, 58).

Additional measures

Team-based approaches to patient management

The importance of the healthcare team in managing patients with sepsis and severe sepsis cannot be overemphasised. Effective management in response to rapid fluctuations in patient health is impossible without highly skilled and devoted healthcare personnel.

Evidence suggests that leadership of critical care teams by full-time, board-certified intensivists improves patient outcomes and reduces patient mortality. Estimates suggest that intensivist-model staffing would save approximately 53,850 lives per year in the US alone (43). 

Education of critical care teams in early patient identification

The rapid provision of appropriate therapies to critically ill sepsis patients is not possible without the availability of highly skilled and experienced critical care teams and outreach personnel. Key to this is the instruction of less experienced staff in the early identification of sepsis patients, both in critical care facilities and on general surgical and medical wards.

Studies from the 1980s have indicated that specialist staff training in critical care medicine reduces mortality in critically ill patients, including those with septic shock (41, 61). The Rivers study (10) (see page 9) provides an excellent example of the mortality reduction benefits of early intervention in patients with severe sepsis and septic shock. Data indicate that delays in the administration of therapies should be avoided in order to reduce mortality rates in sepsis patients.

Goals for mortality reductions in sepsis

Like AMI, effective strategies to combat sepsis are already in place within healthcare systems and, for the first time, new therapies are showing positive results in clinical trials. The Rivers study (10) highlighted the benefits of maintaining patient-specific goals (see page 9). Such goals should be selected carefully to ensure that they are challenging, yet achievable.

But what about the bigger goal? What should that goal be? Should we be arguing for a mortality rate as low as 20% within 10 years, or 10% within 25 years? Will the adoption of such goals help us to structure our efforts of discovery and invention in ways that will speed these advancements?

The setting of such goals will require leadership from national and international societies, and professional bodies, ideally working in collaboration. It will be necessary for these bodies to make appropriate recommendations concerning the management of sepsis, and to drive forward efforts to promote increasing public, professional and governmental awareness of these challenging issues.

Monitoring mortality rates

If we are to set such goals, then periodic observational studies or ongoing patient registries may be necessary to monitor the critical steps of each successive advance from clinical trial to clinical practice.

Most of the available data concerning the impact of new therapies and developments in supportive care, have been derived from carefully conducted and controlled clinical trials. These results cannot always be applied easily to everyday hospital populations, since mortality rates may vary widely, both between countries and between institutions. The increasing application of evidence-based medicine is crucial if effective strategies are to be implemented rapidly.

Many registries are ongoing around the globe to determine the efficacy of patient management strategies in reducing sepsis-associated mortality. These include: ANZICS (Australia/New Zealand); EPISEPSIS  (France); GiViTi (Italy); ICNARC (UK); NICE (The Netherlands); Project Impact (USA); and SICSA (Scotland). In addition, the following registries are currently being established: ERIC (Europe); CCRNet (Canada); SepNet (Germany); The PROGRESS Study (international); and SCCTG (Sweden).  Inclusion of all severely septic patients in these registries is encouraged.

Summary

Sepsis is an increasingly prevalent condition that is associated with a serious risk of irreversible organ dysfunction and death. Numerous highly successful strategies are already in place for the management of sepsis, and additional new therapies are under development. These methods are affording healthcare teams the opportunity to stem, and even reduce, sepsis-associated mortality rates. Unchecked, and in the face of an aging and more at-risk global population, these rates will only rise.

Multifaceted approaches to patient management, the use of evidence-based methods and the adoption of incremental, goal-oriented strategies are vital if we are to combat this complex, aggressive and increasingly prevalent condition. The importance of the whole-hearted involvement of the entire healthcare team and the provision of strong public and political support in achieving these objectives cannot be stressed enough.

References

1. Murray CJL et al. The Global Burden of Disease 2000 in Aging Populations [online] November 2001. Available from URL: http://www.hsph.harvard.edu/burdenofdisease/publications/_papers/gbd2000.pdf
2. American Cancer Society. Cancer Statistics Slide Set [online] 2002. Available from URL: http://www.cancer.org/eprise/main/docroot/PRO/content/PRO_1_1_Cancer_Statistics_2002_slides.zip@sitearea=PR025.
3. Hoyert DL et al. National Vital Statistics Reports [serial online] September 21, 2001. Available from URL: http://www.cdc.gov/nchs/data/nvsr/nvsr49/nvsr49_08.pdf.
4. Angus DC et al. Crit Care Med 2001; 29: 1303–10.
5. Angus DC and Wax RS. Crit Care Med 2001; 29 (Suppl.): S109–16.
6. Balk RA. Crit Care Clin 2000; 16: 179–92.
7. Bone RC. Clin Microbiol Rev 1993; 6: 57–68.
8. Bernard GR et al. N Engl J Med 2001; 344: 699–709.
9. Natanson C et al. Crit Care Med 1998; 26: 1937–31.
10. Rivers E et al. N Engl J Med 2001; 345: 1368–77.
11. Briegel J et al. Clin Invest 1994; 72: 782–7.
12. Bollaert PE et al. Crit Care Med 1998; 26: 645–50.
13. Edbrooke DL et al. Crit Care Med 1999; 27: 1760–7.
14. Rangel-Frausto MS. Infect Dis Clin North Am 1999; 13: 299–312.
15. Brun-Buisson C. Intensive Care Med 2000; 26 (Suppl.): S64–74.
16. Ellis SG et al. Circulation 1994; 90: 2280–4.
17. The Task Force on the Management of Acute Myocardial Infarction of the European Society of Cardiology. Eur Heart J 1996; 17: 43–63.
18. Batchelor WB et al. J Am Coll Cardiol 2001; 38: 1608–13.
19. Roffi M. Circulation 2001; 104: 2767–71.
20. Topol EJ and GUSTO V Investigators. Lancet 2001; 357: 1905–14.
21. International Stroke Trial Collaborative Group. Lancet 1997; 349: 1569–81.
22. Baekgaard N et al. Ugeskr Laeger 1990; 152: 1985–7.
23. Amundsen S et al. Acta Chir Scand 1990; 156: 323–7.
24. Bone RC. Clin Chest Med 1996; 17: 175–81.
25. Friedman G et al. Crit Care Med 1998; 26: 2078–86.
26. Balk RA et al. Sepsis Handbook. National Initiative in Sepsis Education, 2001.
27. Marshall JC. Crit Care Med 2001; 29 (Suppl.): S99–106.
28. Lyseng-Williamson KA and Perry CM. Drugs 2002; 62: 617–30.
29. The GUSTO Investigators. N Engl J Med 1993; 329: 673–82.
30. Daneman N et al. Can J Cardiol 2001; 17: 771–6.
31. Herlitz J. Lakartidningen 2000; 97: 3369–72.
32. Theroux P, Willerson JT, Armstrong PW. Circulation 2000; 102: IV2–13
33. ACC/AHA Pocket Guidelines. The Management of Patients with Acute Myocardial Infarction [online] 2000. Available from URL: http://www.acc.org/clinical/guidelines/nov96/PCKTAMIS.pdf
34. Goebel FD. J Acquir Immune Defic Syndr Hum Retrovirol 1995; 10 (Suppl.1): S62–8.
35. Parker RG. AIDS 1996; 10 (Suppl. 3): S27–31.
36. Matsushita S. Int J Hematol 2000; 72: 20–7.
37. HIV Vaccine Trials Network. The Road to an HIV Vaccine [online] 2001. Available from URL: http://www.scharp.org_3A81/hvtn_pdf/pressroom/theroad.pdf
38. Panel on Clinical Practices for Treatment of HIV Infection. Guidelines for the Use of Antiretroviral Agents in HIV-Infected Adults and Adolescents [online] 2002. Available from URL: http://www.hivatis.org/guidelines/adult/May23_02/AAMay23.pdf
39. UNAIDS. Report on the Global HIV/AIDS Epidemic [online] 2002. Available from URL: http://www.unaids.org/barcelona/presskit/report.html
40. Wheeler AP and Bernard GR. N Engl J Med 1999; 340: 207–14.
41. Reynolds HN et al. JAMA 1988; 260: 3446–50.
42. ARDS Network. N Engl J Med 2000; 342: 1301–8.
43. Young MP and Birkmeyer JD. Effective Clinical Practice [serial online] 2000 Nov–Dec. Available from URL: http://www.acponline.org/journals/ecp/pastiss/nd00.htm
44. Annane D. J Endotoxin Res 7: 305–9.
45. Kollef MH. Crit Care 2001; 5: 189–95.
46. Van den Berghe G et al. N Engl J Med 2001; 345: 1359–67.
47. Coursin DB and Wood KE. JAMA 2002; 287: 236–40.
48. Kellum JA et al. Intensive Care Med 2002; 28: 29–37.
49. Martin GS and Bernard GR. Intensive Care Med 2001; 27 (Suppl.): S63–79.
50. Esteban A et al. J Am Med Ass 2002; 287: 345–55.
51. Vincent JL. Crit Care 2001; 5 (Suppl.): S1–5.
52. Holmes CL et al. Chest 2001; 120: 989–1002.
53. Meier-Hellmann A et al. J Am Soc Nephrol 2001; 12 (Suppl.) S65–9.
54. Kanji S et al. Pharmacotherapy 2001; 21: 1389–402.
55. Healy DP. Ann Pharmacother\ 2002; 36: 648–54.
56. Angus DC. Presented at the Society of Critical Care Medicine’s 30^th International Educational and Scientific Symposium, San Francisco, USA, 2001.
57. Annane D. JAMA 2002; 288: 862-871.
58. Annane D. Minerva Anestesiol 2002; 68: 127–31.
59. Scheen AJ. Rev Med Liege 2001; 56: 859–61.
60. Groeneveld J et al. Crit Care 2002; 6: 102–5.
61. Li TCM et al. JAMA 1984; 252: 2023–7.
62. Jimenez MF and Marshall JC. Intensive Care Med 2001; 27 (Suppl.): S49–62.